Propane aromatization over ZSM-5-based catalysts

This work focuses on the design of catalytic materials for the conversion of the light alkane propane to the value-added products aromatics and hydrogen on ZSM-5-based catalysts using the Discovery Informatics methodology, a modelbased approach of extracting knowledge from data. The development of a robust microkinetic model for this system, involving the study of catalytic descriptors such as Si/Al ratio and percent gallium content, constitutes the primary goal of the work. A microkinetic model using 312 elementary steps and 25 rate and equilibrium parameters to describe the aromatization of propane over HZSM-5 with Si/Al of 16 is the base case against which the effects of Ga and variation of Si/Al are compared. Ga/HZSM-5 catalysts, synthesized by the incipient wetness impregnation technique showed a steady decline in Brønsted acidity with gallium addition. A maximum in propane conversion and aromatics selectivity at a Ga/Al ratio of about 0.5, suggested synergy between proton and gallium sites. Kinetic models based on two different Ga active sites, including GaH2+ and GaH2+ were first used individually to describe the diverse dataset that includes conversion to 10 different products as a function of temperature, space time and Ga/Al variations. An evaluation of these models based on an assigned catalytic functionality for these sites and the associated parameters showed that both sites are required to provide a unified description of the catalytic behavior across gallium content. The ability to discriminate between the models and their implications for the primarily dehydrogenation nature of the Ga active sites will be discussed. A sensitivity analysis of the 25 parameter proton model showed that five primary and four secondary parameters are sufficient to capture the catalysis of propane aromatization on HZSM-5. The resulting nine parameter model, with the remaining sixteen parameters fixed at their estimated values to account for the additional chemical pathways known to be present, is validated with additional data and confidence intervals evaluated by a novel Bayesian estimation approach. We also show that the effect of Ga addition on the nine proton parameters is minor and is focused on competing pathways for hydrogen removal